AI In Medicine From The Patient’s Point Of View
By
Leonard Zwelling
It’s pretty hard to get through a week of The New York Times and The Wall Street Journal without reading a story about the effects of AI on employment. Of late, most of the focus has been the concern for the AI-induced loss of entry level jobs in the white-collar segment of the work force. It is these entry level jobs that provide a large number of college graduates with their first job and their earliest learning about how complex organizations work. These jobs have been the training grounds for tomorrow’s executives and they are vanishing as Claude eats them up like a modern Pacman. Thus, the boos at the commencement addresses by the AI advocates.
It is reasonable to assume that this shortage of entry level, white-collar jobs will hit academic medicine, too. It could include the lower-level jobs in hospital administration, finance, research administration, nursing, academic departmental administration, grant writing and management, and even facilities management.
Throughout my career as a vice president, I had many interactions with young people “learning the ropes” in my own office and with all the offices with which mine interacted. There is little doubt that an AI tool can perform many of the tasks assigned to those in entry level positions and do them faster, better, and cheaper.
Here are just a few I can think of.
Grant writing and review prior to submission: once a faculty member enters the raw information, I suspect Claude can come up with a pretty nifty RO1 application that would be fully referenced, formatted properly, and with an air-tight budget with justification and the necessary increments over the course of the five-year life of the proposed research.
Patient scheduling: I have been through an exasperating experience at MD Anderson trying to get myself cleared by Cardiology for an MRI exam because I now have a pacemaker. The scheduling system at Anderson kept getting things in the wrong order, the MRI before the Cardiology clearance. AI can fix that, no people required.
Nursing: checking patient meds, vital signs, doctors’ orders, and insurance clearances should be a trivial AI task. Why pay people to do all that?
In research administration, something I know a little bit about, tasks like preparing an FDA application for a new drug, reviewing protocol eligibility of patients at the point of service (the clinic), auditing clinical trials, allocating space according to grant dollars, and reviewing issues related to conflict of interest, biosafety and even research misconduct are right in Claude’s wheel house. I could have halved my staff with AI.
So now the worry is what about all the lost jobs? Well, first, this may mean that applicants have to come with better credentials, including AI experience in school. Academic medical institutions may have to establish training programs for new hires so that they are capable of using AI tools to do their jobs, knowing that any new employee may now be able to do the job that used to require three people.
The real payoff is better patient care, more efficient patient management, better administration with fewer people, and readily manageable and comprehensible data analysis to improve operations.
Now the question is, is MD Anderson doing any of this today?
From this patient’s perspective it seems like no one cares enough to get this implemented. Scheduling is a mess. Communicating with care teams is both difficult and always delayed. My Chart is not an efficient way to get to your care team with questions (although it works fine with my Methodist doc). Finally, have you interacted with any person at a clinic front desk of late? Do any of these people seem bright in the least? Friendly? Not at my clinics. And, as an aside, do we really need Jesus statues at every front desk? The Muslim and Jewish patients may find this offensive in this context. I’d rather talk to Claude. He’s an atheist.
The executive leadership of MD Anderson ought to take a good, hard look at the ability of those assigned the tasks relating to bringing MD Anderson into the AI era. To me, as a patient, they seem to be flailing and failing.
Consulting with patients, faculty, and administrative staff with regard to ideas of how AI could improve patient service and work efficiency at Anderson would be a good first step. I know, as a patient, no one has asked me. As I said above, scheduling and most clinic staff leave a lot to be desired.
Finally, it should be obvious that the intelligent implementation of AI in all the areas at Anderson would likely decrease the payroll dramatically leaving far more money for important projects in research and better patient care. I have no idea whether or not AI is on the radar screen of the Board of Regents, but it should be.
